Goelz et al. Brain Informatics (2023) 10:11 Brain Informatics
https://doi.org/10.1186/s40708-023-00190-y
RESEARCH Open Access
Classification of age groups and task
conditions provides additional evidence
for differences in electrophysiological correlates
of inhibitory control across the lifespan
Christian Goelz1, Eva‑Maria Reuter2, Stephanie Fröhlich3, Julian Rudisch3, Ben Godde4, Solveig Vieluf1,5 and
Claudia Voelcker‑Rehage3*
Abstract
The aim of this study was to extend previous findings on selective attention over a lifetime using machine learning
procedures. By decoding group membership and stimulus type, we aimed to study differences in the neural represen‑
tation of inhibitory control across age groups at a single‑trial level. We re‑analyzed data from 211 subjects from six age
groups between 8 and 83 years of age. Based on single‑trial EEG recordings during a flanker task, we used support
vector machines to predict the age group as well as to determine the presented stimulus type (i.e., congruent, or
incongruent stimulus). The classification of group membership was highly above chance level (accuracy: 55%, chance
level: 17%). Early EEG responses were found to play an important role, and a grouped pattern of classification perfor‑
mance emerged corresponding to age structure. There was a clear cluster of individuals after retirement, i.e., misclas‑
sifications mostly occurred within this cluster. The stimulus type could be classified above chance level in ~ 95% of
subjects. We identified time windows relevant for classification performance that are discussed in the context of early
visual attention and conflict processing. In children and older adults, a high variability and latency of these time win‑
dows were found. We were able to demonstrate differences in neuronal dynamics at the level of individual trials. Our
analysis was sensitive to mapping gross changes, e.g., at retirement age, and to differentiating components of visual
attention across age groups, adding value for the diagnosis of cognitive status across the lifespan. Overall, the results
highlight the use of machine learning in the study of brain activity over a lifetime.
Keywords Machine learning, Decoding, EEG/ERP, Flanker, Selective attention, Development
*Correspondence:
Claudia Voelcker‑Rehage
claudia.voelcker‑rehage@uni‑muenster.de
Full list of author information is available at the end of the article
© The Author(s) 2023. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which
permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the
original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or
other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line
to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory
regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this
licence, visit http://creativecommons.org/licenses/by/4.0/.
, Goelz et al. Brain Informatics (2023) 10:11 Page 2 of 11
Graphical Abstract
Classification of age groups and task conditions provides additional evidence for differences in electrophysiological
correlates of inhibitory control across the lifespan
Methods and participants Results
1. Classification of age groups: 2. Classification of trial type:
Participants: N=222, age: 8-86 years, 5 age groups (children - old
• confusion matrix reveals clusters • differences in latency of
Flanker task adults) Trial type which may represent groups with maximum performance
(adopted from Winneke et al., 2019)
similar with similar cognitive status between different age groups
Classification of trial type: time resolved decoding performance
Inconcucruent
200 ms
Congruent
Pe r f or m a nce [ AUC]
300 ms 0.7
~950 ms
200 ms 0.6
200 ms
300 ms 0.5
0.0
EEG Classification
0.0 0.2 0.4 0.6
1. Between participant: Tim e [ s]
Ch. 1 age groups
Ch. 2 Conclusion
.
. 2. Within participant: Our analysis was sensitive to mapping gross changes, e.g., at retirement age, and to differentiating
.
Ch. 32 xDAWN + SVM trial type components of visual attention across age groups. Machine learning thus extends classical
methods in that it can be used to test existing theories but also to extend them.
1 Introduction Recently, the use of machine learning techniques
Selective attention (as part of inhibitory control) to study experimental effects in EEG studies gained
describes the ability to focus on relevant information popularity as a complement to classical ERP analyses.
while simultaneously suppressing irrelevant or distract- These methods are referred to as multivariate pattern
ing input which is essential for the accomplishment analysis (MVPA) or decoding approaches and are based
of complex tasks [1]. This ability changes throughout on classification algorithms developed in the field of
the lifespan. Whereas selective attention develops in brain–computer interfaces (BCI) [5]. The main idea is
children due to the differentiation of brain areas and to train a machine learning model based on single-trial
networks, the opposite is noticeable in older adults, EEG data that allows to classify a certain behavior or
namely, a reduction in selective attention likely related experimental condition. This involves the automatic
to dedifferentiation processes in the brain [2, 3]. Focus- detection of generalizable multivariate patterns associ-
ing on the neuronal response to sensory stimuli meas- ated with the behavior or experimental condition. Tar-
ured with electroencephalography (EEG) differences in geting the information content on a single-trial level
the distribution, amplitude, and latency of event-related with respect to an experimental condition rather than
potentials (ERP) were reported for different age groups averaged activation on single electrodes and time win-
[3]. Comparing six different age groups Reuter et al. dows, such approaches can be seen as complementary
[4] confirmed a u-shaped function of ERP markers for to classical univariate ERP analyses [5]. Classification
encoding and processing speed (i.e., P1, N1, N2, and P3 approaches are less dependent on a priori assumptions
latencies), markers of visual processing and attention (e.g., selection of electrodes or time windows), and
(i.e., P1 and N1 amplitudes) as well as gradual changes naturally simplify the problem of multiple comparisons
in markers of cognitive processing (N2, P3 amplitudes, [6]. In this way, these methods have higher sensitivity
and P3 distribution). Moreover, results suggest that by exploiting the interdependence of EEG signals while
different neural mechanisms underly performance in omitting the information loss due to trial averaging.
children and older adults [4]. The u-shaped function in Moreover, additional analyses allow characterizing the
previous ERP findings suggests that ERP components cortical representation of a high number of stimuli, e.g.,
are similar between children and older adults despite their dynamics and similarity [5, 7]. Nevertheless, these
fundamentally different mechanisms, i.e., differentia- methods should be considered as a complement to the
tion in children versus dedifferentiation in older adults. classical univariate methods, since directional effects
It is unclear whether these differences are reflected in cannot be represented. To study the neuronal response
distinctive brain activation patterns and to what extent to sensory stimuli on single-trial level decoding
lifelong changes in electrophysiological markers can be approaches are often used in a time-resolved manner
detected at the level of individual trials. to investigate the time course of information density in
relation to the stimulus. This includes determining the
